Chemical Signals in the Marine Environment: Dispersal, Detection, and Temporal Signal Analysis
Chemical signals identify biologically important targets for those who have the proper receivers. We assume that selection pressure can act on both the biochemical and the physiological regulation of the signal and on the morphological and neurophysiological filter properties of the receiver. Communication is implied when signal and receiver evolve toward more and more specific matching, culminating in well-known sex pheromone systems. In other cases, receivers respond to portions of a body odor bouquet that is released to the environment not as a (intentional) signal but as an unavoidable consequence of metabolic activity or tissue damage. Breath, sweat, urine, feces, their aquatic equivalents, and their bacterial and other symbiotic embellishments all can serve as identifiers for chemoreceptive animals interested in finding food or hosts. Body fluids released from damaged tissues and decay products from dead organisms can be particularly potent signals. Since all organisms must release metabolites in order to live, and since any such release is a potential target of opportunity for predators and parasites, one may expect that several forms of chemical camouflage have evolved to obscure one's chemical presence. Both communication signals and camouflage depend on signal-to-background contrast or lack thereof: communication emphasizes contrast; camouflage works toward lower contrast.
Jelle Atema is professor of biology and director of the Boston University Marine Program at the Marine Biological Laboratory, Woods Hole, Massachusetts.
Contrast can be provided by chemical specificity of the signal (spectral contrast) and by temporal changes in concentration of the signal (dynamic temporal contrast). Spectral contrast is created by unique compounds and by unique mixtures of compounds, including ordinary ones. Temporal contrast emerges from the rate at which the concentration of a compound changes with time, including the repetition rate. Temporal changes reflect spatial patchiness and hold information for chemotactic behavior at different spatiotemporal scales. The two classical methods of camouflage, well-known in the visual signal world, may also operate in the chemical signal world, although they are virtually unstudied. To avoid detection, animals with visual predators hide and remain motionless, or they look and move like their background; animals with chemically hunting predators may build impermeable shells and store urine and feces until it is safe to release them, or they may produce metabolites that match the environment in mixture composition and temporal distribution.
Unlike wave or wave-like propagation of acoustic, visual, and other electromagnetic signals, chemical signals disperse through the environment by molecular diffusion and bulk flow. At small spatial scales—in practice below 10 µm—diffusion is a biologically useful transport mechanism and, given the constraints of viscous fluid boundary layers, often the only effective mechanism. At larger scales, flow is necessary to obtain metabolic energy (e.g., oxygen, food particles), to eliminate wastes (e.g., carbon dioxide, urine), and to send and receive chemical signals. The constraints of metabolism and sensory information are probably different, so that we could expect animals to generate separate metabolic currents and information currents. In practice, they may use the same current-generating mechanisms and then control the currents and the chemical composition of these currents to serve different functions at different times. Controlling the timing, velocity, and direction of information currents is important whether they are used to send or receive chemical signals. Animal-generated currents can be laminar at small scales (<1 cm) or turbulent at larger scales. Both include the possibility of temporal information. In this paper I will focus on temporal information in marine chemical signals and on the use of urine dispersal in chemical communication.
The marine environment is filled with sources of chemical signals in a wide range of overlapping spatial scales (1), from the metabolites of a single marine bacterium (diameter, <10-6 m) to the odor plumes left behind a traveling school of tuna (school size, >102 m) or emanating from a whale carcass (plume size, >103 m). Constrained by physics, chemistry, and biology, chemical signals have a finite lifetime. When released into the environment, they disappear below detectable levels as a result of turbulent mixing, molecular diffusion, adsorption, photolysis, and chemical transformation and through uptake and breakdown by bacteria,
microorganisms, small invertebrates, and invertebrate larvae. During their lifetime, chemical signals exist as patches of constantly varying sizes and shapes. Temporal scales tend to follow spatial scales: molecular diffusion would obliterate a 10-µm (diameter) spherical patch of small dissolved molecules in 10 s (i.e., concentration drop to 0.1% of original) and a 1000-m-long surface oil slick resulting from a whale carcass being devoured by sharks can be visible for days.
Since all organisms during their lifetime experience only a limited range of spatiotemporal scales, it is important to understand the scales that are relevant to particular animals if we are to discover the mechanisms animals use to extract information from the dispersal patterns of chemical signals. Signal longevity and temporal pattern are no less important in chemical signals than they are in other sensory stimuli. The speed of temporal signal detection and processing depends on the encounter rate with odor patches and their spatial gradients. This encounter rate is determined by the speed of search behavior. Perhaps as a result of this correlation we tend to see a dependence on chemical signals in "slow" animals—e.g., crustacea, mollusca (but not visually hunting squid). Needless to say, many slow animals can have very fast escape responses, thus linking their slow search to sensory processing, not to intrinsic locomotion limitations. In lobsters, chemical gradient search reduced their normal walking speed by half, and their chemoreceptive flicker fusion rate (i.e., the stimulus pulse frequency that causes responses to fuse) is an order of magnitude lower than typical visual fusion rates.
In order to detect chemical signals in the sea, receptors must recognize them against a background of many other chemical compounds. This involves not only receptor specificity and diversity but also recognition of the intensity and time course of the signal to allow a receptor to distinguish a true signal from random events. A point receptor moving across a field of patches will see them as a chaotically fluctuating intensity pattern. We assume therefore that receptors evolved with both spectral and temporal properties tuned to signal recognition and that the tuning properties of receptor organs reflect, on the one hand, the constraints imposed by prevalent natural stimulus conditions and, on the other hand, the demands for specific information most useful for that animal's behavioral tasks. In a true sense, the receptor organ determines by its spectral and temporal tuning what is signal and what is noise. The same environmental signal distribution can therefore yield different information as a result of tuning. This applies to different species as well as to different organs of a single species. Note that many species have a multitude of different chemoreceptor organs, mostly unstudied physiologically and behaviorally. By comparison, humans have three, perhaps
four, chemoreceptor organs serving the senses of smell, taste, trigeminal, and perhaps vomeronasal chemoreception, each tuned to a different spectrum of stimuli and to different encounter rates.
A full understanding of chemical ecology must therefore include not only the characterization of chemically unique signals but also their environmental dispersal and degradation patterns that are an intrinsic part of chemosensory transduction and signal processing and lead to the appropriate behavioral responses. One of the most difficult tasks has been to measure the natural stimulus dispersal patterns at a spatial and temporal resolution relevant to the animal.
In this paper I will discuss a single species to illustrate some of the processes of marine chemical signal dispersal, receptor tuning, and chemotactic and social behavior. In an effort to understand the underwater world of chemical signals—so foreign to humans—we attempt to see the marine environment through the many chemosensory organs of the lobster, Homarus americanus, an animal that has demonstrated its ability to communicate with chemical signals, urine release, and a variety of information currents and to extract spectral and temporal chemical information from its turbulent environment (2). To investigate the spatiotemporal dynamics of natural odor dispersal that are important physiologically and behaviorally, aquatic chemical signals and large arthropods offer significant advantages. Due to the greater density of water, the relevant scales of turbulence are about an order of magnitude smaller in water than in air, facilitating behavioral experiments and odor dispersal measurements. Chemical signals in solution can be quantified and expressed in molarity. Electrochemical microelectrodes can be constructed to measure marine chemical stimulus patterns at the proper spatial scale of receptor sensilla, and recent advances in signal processing have allowed the temporal resolution to exceed the animal's (3). These electrodes mimic crustacean chemoreceptor organs, which are built with cuticular pegs that form distinct boundaries with the surrounding fluid, uncomplicated by mucus transitions. Large arthropods can easily carry instrumentation packets, including catheters, electrodes, amplifiers, and transmitters.
SIGNAL RECEPTION AND ANALYSIS: TEMPORAL ANALYSIS OF CHEMICAL SIGNALS TO DETERMINE SPATIAL GRADIENTS
This section reviews four different experimental approaches that together argue in favor of a temporal analysis function of lobster olfaction. The experiments include high-resolution measurements of turbulent odor dispersal and lobster sampling behavior, electrophysiological recording of in situ single cell responses to controlled and chaotic stimuli, and behavioral analysis of orientation and localization of odor sources.
Spatial Gradients in Turbulent Odor Plumes
Since turbulent odor plumes simultaneously contain a broad suite of eddy sizes, the first problem is to decide which range of frequencies to measure and at which spatiotemporal scale to resolve the signal/noise complex. For spatial resolution, we scaled our odor tracer sensor (carbonfilled glass microelectrode) to the physical dimension of the lobster's 30-µm (diameter) olfactory receptor sensillum. For temporal resolution, we chose 5 ms [IVEC-10 (in vivo electrochemistry analyzer), and the Gerhardt custom package, Medical Systems Corp., Greenvale, NY], exceeding both the lobster's olfactory flicker-fusion frequency of 4 Hz (see below) and laboratory plume frequencies, which have little energy above 40 Hz. We studied odor dispersal patterns resulting from biologically scaled, constantly emitting jet sources in slow background flow. From high-resolution plume measurements using dopamine as a tracer, we described the spatial distribution of encounter probabilities of eddy features such as peak concentration, concentration gradients at their leading edge, intermittency, etc. (4). Some of these stimulus features showed spatial gradients that could be used to track and locate an odor source. This demonstrates that at the measuring scale of animal sensors, purely chemoreceptive information can be extracted from which the direction of a distant odor source can be estimated. In other words, true chemotaxis based on temporal analysis of odor patch features is theoretically possible.
Orientation and Navigation in Odor Plumes
Subsequent behavioral experiments in the same laboratory flume (5) showed that lobsters indeed located the source of food odor plumes, whereas prior lesion experiments had shown that distance orientation is guided by olfaction (6) and the final approach of about 30 cm by taste (7). Smell and taste are defined functionally (8). The olfactory orientation was composed of an initial scanning phase (characterized by low walking speed and large heading angles) followed by an increasingly fast and accurate approach. Both speed and accuracy reached a maximum that was maintained until walking legs (taste) took over the final search. Curiously, when not engaged in chemotactic search, lobsters can walk twice as fast. Unlike some insects (9, 10), lobsters do not appear to use innate motor programs, such as counterturning. Instead, the individuality of their tracks and their relatively slow top speed during chemotactic approach suggest that they monitor and follow the variable spatial gradients characteristic of turbulent plumes. If walking animals such as lobsters use odor patch information for plume navigation, then free-
swimming animals without ground reference will be even more likely to extract directional cues from turbulent odor dispersal patterns.
Two questions arise from this result. Do lobsters use only chemical and not mechanosensory information, and why do lobsters not use ground reference and head up-current? Since turbulent odor dispersal is based on water flow patterns, we must investigate the role of microflow patterns in plume orientation behavior. As for ground reference, we speculate that the flow patterns of the lobster's natural environment may be too complex to allow for efficient rheotactic behavior in odor source localization. This complexity is most likely caused by a mismatch between turbulent scales and animal body size and sampling scales.
Odor Sampling Behavior
Feature extraction is the primary function of sensory systems. To this end, sense organs almost always use three different levels of signal filtering: a physical filter based on receptor organ morphology and associated behavior, a receptor cell filter based on biophysical and biochemical properties of transduction and adaptation, and a neural filter based on network connectivity in the central nervous system. We investigated the first two filters for lobster olfaction. The olfactory sensilla (''hairs") known to be critical for efficient orientation behavior (6) form a dense "toothbrush" on the distal half of the lateral flagellum of the antennules. Video analysis and high-resolution electrochemical measurements showed that under low flow conditions (<5 cm/s) this brush forms a dense boundary layer which traps existing odor and shields the receptors from rapid odor access, thus making smelling virtually impossible (3). Thus, lobsters must flick their antennules (i.e., sniff) to smell. Flicking behavior drives water at high velocity (>12 cm/s) through the brush and causes the hairs to tremble in their sockets. This allows rapid odor exchange around the entire 1-mm-long shafts of all the hairs (3). Flick rates of up to 4 Hz occur in excited lobsters (11). The bilateral antennules allow for spatial comparison essential for efficient orientation (6).
Temporal Resolution of Olfactory Receptor Cells
Lobster chemoreceptor cells show a great diversity of filter properties. Electrophysiological measurements show that each receptor cell is tuned not only to one or a few preferred compounds (12, 13) but also to a preferred frequency (14, 15). Temporal resolution in chemoreception is affected by at least five different stimulus parameters: rate of stimulus concentration increase (= pulse slope), amplitude and duration of a
single odor pulse, pulse repetition rate, and pulse-to-background concentration ratio. Pulse slope corresponds to the arrival of an odor-flavored eddy, duration corresponds to its size; repetition rate represents arrival of different eddies and their spacing. All five stimulus parameters depend on the degree of turbulent and diffusive mixing and thus correlate with time since release and distance from the source. The dynamic response properties of receptor cells are determined by two somewhat independent cellular processes: adaptation and disadaptation (or recovery). A cell's adaptation rate determines its preferred stimulus slope and duration, and its recovery rate determines its flicker fusion frequency. Results from qualitative experiments in which odor and single cell responses were measured simultaneously and with high spatiotemporal resolution indicate that steep pulse slope and large interpulse interval are important excitatory stimulus features. Unfortunately, the nonlinear dynamics of adaptation and disadaptation processes preclude a simple solution for determining meaningful transfer functions. Therefore, stimulus features must be analyzed one at a time to measure their effect on receptor cell responses, and—ultimately—to reconstruct the temporal analysis capabilities of the lobster olfactory organ.
Carefully controlled square pulses of odor with identical amplitude and different pulse duration (50-1000 ms) resulted in response maxima for 200 ms stimuli; longer pulse durations did not result in greater responses as measured by spike firing frequency. This reflects the effects of cellular adaptation mechanisms-presumably receptor phosphorylation-that begin to overwhelm excitatory processes at about 200 ms. For all stimulus pulse amplitudes, adaptation is complete at about 400 ms. Thus, responses to a single pulse are determined largely by initial stimulus pulse parameters and by the cell's intrinsic physiological properties and negligibly by pulse features beyond 200 ms. Flicker fusion to 100 ms square pulses occurs at 4 Hz for the fastest receptor cells (15). Some insect receptor cells can follow 4 to 10 Hz pulse rates (16-18). These results indicate that cellular disadaptation processes require at least 150 ms before the cell can respond again to phase-lock with subsequent pulses of similar magnitude. In addition, pulse amplitude-to-background ratio determines response magnitude: backgrounds shift the response function to pulses toward higher concentrations.
The results demonstrate (i) that in turbulent odor plumes, purely chemical spatial gradients can be calculated when measuring with sensors scaled to lobster olfactory organs, (ii) that rapid odor access to the lobster's olfactory organs (under low ambient flow conditions) is accom-
plished by flicking, (iii) that the maximum observed flick rate of 4 Hz corresponds to the neurophysiologically determined flicker fusion frequency of olfactory receptor cells, (iv) that receptor cells are tuned not only to specific compounds but also to different temporal parameters of odor, and (v) that lobsters locate odor sources with search paths, walking speed, and turning behavior suggestive of chemotaxis based on patchy odor distributions. These results show that turbulently dispersed odor patches from a single source contain spatial gradients of signal parameters and that some of these parameters are recognized by olfactory temporal filters. We hypothesize that these temporal odor parameters provide information for chemotactic navigation in odor plumes. At this point, we have learned some of the physiologically determined temporal filters, but we do not know the exact nature of the behaviorally relevant signal features nor the sampling regimes and signal processing required to lead to efficient navigation. We are approaching these questions with behavioral, physiological, computational (19), and robotic methods (20). In the context of pheromone communication, Bossert (21) calculated that amplitude modulation could enhance information transmission very significantly. Some moths release pheromone in puffs with 1-s periodicity (22) and it was found that Bossert's calculations must be reevaluated in light of current knowledge of temporal filter properties of chemoreceptor cells and the unpredictability factor of turbulent dispersal that characterizes most natural environments.
SIGNAL PRODUCTION AND BROADCASTING: URINE DISPERSAL IN CHEMICAL COMMUNICATION
This section reviews the complex currents lobsters generate to eliminate metabolites and broadcast chemical signals and the return currents from which they obtain chemical signals and metabolic energy. Lobsters are examples of hard-shelled animals that store urine and feces, allowing them to be chemically "quiet" when necessary.
Information Currents and Urine Signals
H. americanus utilizes three current-generating mechanisms that can operate separately or in combination; all three are implicated in chemical communication. The scaphognatites inside the gill chambers generate a powerful gill current which jets forward from bilateral "nozzles." This current reaches distances of up to seven body lengths in adults (2) and velocities of 3 cm/s near the nozzle. It is usually a bilateral current and it carries the animal's gill metabolites. Mature-sized lobsters under summer temperatures rarely cease producing this breathing current; in winter, the
current stops for episodes of several seconds, presumably reflecting the animal's lower metabolic demands. In addition, urine can be released into this current from bilateral bladders through small ventrally directed nephropores at the base of the antennae. Lobsters appear to release into the urine the products of a small cluster of glands located in the nephropore nipple some 100 µm inside the excretory pore (23). Both the gland with its duct and surrounding muscle tissue and the muscular valve of the nephropore appear designed to give the animal control over chemical signals released into the gill current. Glandular products and function are unknown, but morphology suggests that it is a rosette gland related to common tegumental glands (24), and histological stains suggest that its product is proteinaceous, carbohydrate, or glycoprotein (25).
Further control of signaling is possible through a redirecting of the gill current by the exopodites of the three maxillipeds. It appears that the exopodite of the first maxilliped can be positioned to—partially—cover the gill chamber outflow nozzle, thus deflecting and redirecting forward water flow. The large feathery exopodites of the second and third maxillipeds then fan the deflected water backwards, while drawing in a slow flow of water from around the animal's head (2). The antennules flick and thus sample odor within this area, the radius of which is about the length of the antennule. The exopodite fan current represents the second lobster-generated current. It can be bilateral or unilateral on either side.
Together, the two lobster-generated currents that can be measured around the animal's anterior end are complex and carefully controlled. They are ideally suited to carry urine, urine pheromones, and gill metabolites away from the lobster to specified directions. Simultaneously, the water displaced by these outgoing currents results in incoming currents with chemical signals from the environment that can be sampled by the antennular chemoreceptors.
The third and most powerful lobster-generated current is the pleopod current, which draws water from below the lobster and blows it posteriorly (2). Typically, the lobster raises its tail and beats its pleopods to generate this current, which is sufficiently powerful in adult animals to help the animal in forward motion and in climbing onto rocks. In smaller animals, particularly in IVth stage post-larvae, this current is used for forward swimming and food particle capture (26). Cohabiting mature males pleopod-fan frequently at the entrance of their shelter, thereby sending male odor and the odor of the cohabiting female into the environment. In the laboratory, this behavior results in greatly increased visits to his shelter by premolt females and other lobsters (27). The signals involved are unknown but could easily include urine-carried pheromones.
Urine Signals in Dominance and Courtship
Males compete for dominance, which they establish during fights. Experiments with chronic catheters show that aggressive animals release pulses of urine during the fight and that lobsters stop urine release as soon as they have lost (28, 29). Urine release was not observed when lobsters were disturbed by the experimenter, thus implying that urine release by aggressive animals is social and carefully controlled (28). During subsequent encounters, as long as a week (but less than 2 weeks) later, the opponents remember each other, as evidenced by the fact that almost no fight takes place: the former loser avoids the winner (30). Chemical signals in the urine are involved in this memory, as demonstrated by experimental evidence that both catheterization of urine release and temporary lesion of antennular chemoreceptors result in renewed fights on subsequent days (29, 30).
After dominance is established, the dominant male occupies a "preferred" large shelter which becomes a focus of social interactions. Mature, premolt females visit frequently (31, 32). For cohabitation and subsequent mating, females in naturalistic aquaria chose the dominant male over subdominants (31). In choice tests females prefer larger dominant males (23). Females make these behavioral decisions both from a distance and at the shelter entrance. Discrimination by females is lost when males are catheterized and can be regained when that male's urine is artificially released near the male (23). Thus, male urine cues for female choice are implied but have not been identified.
Visiting females stand still at the entrance of the male shelter for many seconds. Observation of their expodites shows that they alternate between fanning and not fanning. Therefore, when not fanning, they blow their gill current into the male shelter. At the time of female visits, the male often stands inside and away from the female entrance, flicking his antennules, fanning his exopodites (thus drawing water toward his antennules and redirecting his gill current backwards), and occasionally fanning his pleopods—all resulting in female odor reaching the male. In an agonistic context, field (32) and laboratory observations show that a female has to be considerably larger to force a male out of his shelter. In a courtship context, males accept even much smaller premolt females and cohabit with them, one at a time, each for 1-2 weeks. Although such context appears to be provided by chemical signals (33, 34), female sex pheromones have not been chemically identified. We speculate that special glandular products in the urine may be involved to identify the visiting female as a mature and premolt female lobster. Since both males and females at all molt stages possess active nephropore glands (23), its product would have to be modified for different sexes and state of
maturity if it were to be used in the contexts described above. Alternatively, changes in perception of the signal could be involved.
Cohabitation lasts from a few days to weeks. The female molts during this time, and mating follows the female molt after one-half hour. Over the premolt cohabitation period, male pleopod fanning increases. It reaches its maximum during the days of female molting and wanes in the postmolt cohabitation days. Fanning males often stand at one of their shelter entrances with their abdomen raised high and slightly outside the entrance. This results in a strong current running through and out of his shelter (35). This current contains all male and female metabolites (including possible pheromones) released by the cohabiting pair. Male fanning is positively correlated with visits of other lobsters, including premolt females, to the shelter. These other premolt females do not molt but wait for their turn to cohabit with the dominant male (27). Female molt staggering implies control of the female molt cycle. We speculate that the male pleopod current, which serves as an advertisement device to attract lobsters ("releaser pheromones;" ref. 36), also contains molt inhibiting signals ("primer pheromones;" ref. 36) for visiting females who are released from this inhibition when they start cohabiting with the dominant male leading to molting and mating within days.
All indications are that we are only just beginning to see a few threads of the rich fabric of chemical signals that link lobsters to each other and to their environment. Exoskeleton, bladders, glands, and control of currents all indicate that these animals can be chemically quiet and release specific signals at critical times during aggression and courtship. Chemical signals appear to be used to remember individuals and to facilitate stable dominance hierarchies.
One may wonder why lobsters appear to use urine as a dispersal solvent for chemical signals, whereas terrestrial arthropods such as the well-studied insects use direct release of gland products into the air. Perhaps the answer is that small animals in air (such as insects) are always in danger of desiccation. By contrast, marine lobsters and crabs are relatively large and may experience only minor water loss problems due to osmosis. Thus, it may not be difficult for a 500-g lobster to store 10 ml of urine and release it during a dominance battle at a rate of up to 1 ml/min (27). The advantage of urine-carried pheromones is that the dispersal mechanism already exists: urine is injected into the gill current, which in turn injects into ocean currents.
In conclusion, lobsters are excellent models to learn about the natural world of odor dynamics, but they are probably not unique: they are
opening our minds to possibilities not yet explored with other animals. The odor environment is richer and more complex than we know today and lobsters are showing us several thus far unexplored dimensions: navigation based on the characteristics of rapid odor signals, information currents, urine-based pheromone broadcasting in the sea, and—possibly—chemical camouflage.
Chemical signals connect most of life's processes, including interorganismal relationships. Detection of chemical signals involves not only recognition of a spectrum of unique compounds or mixtures of compounds but also their spatial and temporal distribution. Both spectral and temporal signal processing determine what is a signal and what is background noise. Each animal extracts its unique information from the chemical world and uniquely contributes to it. Lobsters have provided important information on temporal signal processing. Marine chemical signals can be measured with high spatio-temporal resolution giving us a novel view of the lobster's environment. Lobster chemoreceptor cells have flicker fusion frequencies of 4 Hz and can integrate stimuli over 200 ms, closely corresponding to odor sampling behavior with 4 Hz "sniffs." Using this information, spatial odor gradients can be determined from temporal analysis of odor patches typical of turbulent dispersal. Lobsters appear to use this information to locate odor sources. Lobster social behavior depends greatly on chemical signals. Urine carries important information for courtship, dominance, and individual recognition. A novel gland in the nephropore is strategically located to release its products into the urine. Urine, in turn, is injected into the gill current, which jets water 1-2 m ahead of the animal. Lobsters control three different currents that carry chemical signals to and from them. The study of odor dynamics has only just begun. It will be exciting to see how signal dispersal, receptor temporal tuning, neural processing, and animal behavior interact to enhance signals for communication and detection and to reduce signals for chemical camouflage.
I thank my students, postdoctoral associates, and colleagues for their many contributions to the exciting discoveries made in the chemical world of lobsters. For recent contributions I mention specifically my long-time colleague Dr. Rainer Voigt, my former students Dr. Paul Moore and Dr. George Gomez, my postdoctoral associates Drs. Thomas Breithaupt and Jennifer Basil, and my students Paul Bushmann and Christy Karavanich. This work was supported by National Science Foundation Grants IBN-9212650 and IBN-9222774 and National Institutes of Health Grant 5 PO1NS25915.
1. Atema, J. (1988) in Sensory Biology of Aquatic Animals, eds. Atema, J., Popper, A. N., Fay, R. R. & Tavolga, W. N. (Springer, New York), pp. 29-56.
2. Atema, J. (1985) Soc. Exp. Biol. Symp. Ser. 39, 387-423.
3. Moore, P. A., Atema, J. & Gerhardt, G. A. (1992) Chem. Senses 16, 663-674.
4. Moore, P. A. & Atema, J. (1991) Biol. Bull. 181, 408-418.
5. Moore, P. A., Scholz, N. & Atema, J. (1991) J. Chem. Ecol. 17, 1293-1307.
6. Devine, D. V. & Atema, J. (1982) Biol. Bull. 163, 144-153.
7. Derby, C. D. & Atema, J. (1982) J. Exp. Biol. 98, 317-327.
8. Atema, J. (1977) in Olfaction and Taste VI, eds. Le Magnen, J. & MacLeod, P. (Information Retrieval, London), pp. 165-174.
9. Akers, R. P. (1989) J. Chem. Ecol. 15, 183-208.
10. David, C. T. & Kennedy, J. S. (1987) Naturwissenschaften 74, 194-196.
11. Berg, K., Voigt, R. & Atema, J. (1992) Biol. Bull. (Woods Hole, Mass.) 183, 377-378.
12. Voigt, R. & Atema, J. (1992) J. Comp. Physiol. A 171, 673-683.
13. Voigt, R. & Atema, J. (1994) in Olfaction and Taste XI, eds. Kurihara, K., Suzuki, N. & Ogawa, H. (Springer, New York), p. 787.
14. Gomez, G., Voigt, R. & Atema, J. (1994) J. Comp. Physiol. A 174 , 803-811.
15. Gomez, G., Voigt, R. & Atema, J. (1994) in Olfaction and Taste XI, eds. Kurihara, K., Suzuki, N. & Ogawa, H. (Springer, New York), pp. 788-789.
16. Marion-Poll, F. & Tobin, T. R. (1992) J. Comp. Physiol. A 171, 505-512.
17. Rumbo, E. & Kaissling, K.-E. (1989) J. Comp. Physiol. A 165, 281-291.
18. Christensen, T. & Hildebrand, J. (1988) Chem. Senses 13, 123-130.
19. Mountain, D. & Atema, J. (1993) Neurosci. Abstr. 19, 121.
20. Consi, T. R., Atema, J., Goudey, C. A., Cho, J. & Chryssostomidis, C. (1994) Proc. IEEE Symp. Auton. Underwater Vehicle Tech., 450-455.
21. Bossert, W. H. (1968) J. Theoret. Biol. 18, 157-170.
22. Connor, W. E., Eisner, T., Vander Meer, R. K., Guerrero, A., Ghiringelli, D. & Meinwald, J. (1980) Behav. Ecol. Sociobiol. 7, 55-63.
23. Bushmann, P. & Atema, J. (1993) Biol. Bull. (Woods Hole, Mass.) 185, 319-320.
24. Harrison, F. W. & Humes, A. G., eds. (1992) Microscopic Anatomy of Invertebrates (Wiley, New York), Vol. 10.
25. Bushmann, P. & Atema, J. (1994) Chem. Senses 19, 448.
26. Lavalli, K. & Barshaw, D. (1989) Mar. Behav. Physiol. 15, 255-264.
27. Cowan, D. F. & Atema, J. (1990) Anim. Behav. 39, 1199-1206.
28. Breithaupt, T. & Atema, J. (1993) Biol. Bull. (Woods Hole, Mass.) 185, 318.
29. Breithaupt, T., Karavanich, C. & Atema, J. (1994) Chem. Senses 19, 446-447.
30. Karavanich, C. & Atema, J. (1991) Biol. Bull. (Woods Hole, Mass.) 181, 359-360.
31. Atema, J., Jacobson, S., Karnofsky, E., Oleszko-Szuts, A. & Stein, L. (1979) Mar. Behav. Physiol. 6, 227-296.
32. Atema, J., Borroni, P., Johnson, B. R., Voigt, R. & Handrich, L. S. (1989) in Perception of Complex Smells and Tastes, eds. Laing, D. L., Cain, W., McBride, R. & Ache, B. W. (Academic, Sydney), pp. 83-100.
33. Atema, J. & Engstrom, D. G. (1971) Nature (London) 232, 261-263.
34. McLeese, D. W. (1973) J. Fish. Res. Board Can. 30, 775-778.
35. Atema, J. (1986) Can. J. Fish. Aquat. Sci. 43, 2283-2290.
36. Wilson, E. O. & Bossert, W. H. (1963) Recent Prog. Horm. Res. 19, 673-716.